Swarm Intelligent Optimization Approach For Dynamic Industry Production Process

With more harsh requirements in environmental and conditional control in industry production process,the industry optimization embodies the complex features of nonlinearity,uncertainty and high latitude.Dynamic optimization method is exactly an effective and popular way to deal with these difficulties and brings in greater economic benefits.Aimed at handling control systems dynamically changing over time,dynamic optimization has been widely utilized in various fields like industry production,navigation and machine manufacture.Control vector parameterization(CVP)is a commonly utilized approach in numerical analysis.The control variables are approximated as a set of variables by a group of base functions in time domain by CVP.Thus,the standard dynamic optimization problem is transformed into a nonlinear programming(NLP)problem with limited parameters in mathematics.Swarm intelligence based evolutionary computation approach has gradually been playing a more and more important role in optimal control cause it not only is simple and easy to implement by programming but also is computationally distributed and has strong robustness.This thesis mainly studies the applications of CVP based swarm intelligent approaches for dynamic optimization problem in industry production process.The main contributions and work of the thesis can be summarized as follows:(1)A general algorithm structure to solve dynamic optimization problems by using swarm intelligent algorithm is proposed.Firstly,the control variables are represented as a set of base functions with CVP.Thus,the original infinite dimensional control problem is transformed into a mathematical programming problem,to which swarm intelligent algorithm is applied in order to obtain approximate optimal control trajectory.(2)To verify the validation of the above algorithm structure,three swarm intelligent approaches(invasive weed optimization(IWO),flower pollination algorithm(FPA)and firefly algorithm(FA))are applied in dynamic optimization problems on the basis of CVP approach.The research results of several classic chemical dynamic optimization problems show the effectiveness of the three proposed CVP based intelligent approaches.(3)Given that swarm intelligent algorithms are tend to be trapped in local optimum when dealing with dynamic optimization problems with strong nonlinearity,two new improved algorithms,named adaptive dispersion IWO based approach(ADIWO-CVP)and adaptive FPA based approach(AFPA-CVP)are further proposed to maintain the exploration ability of the algorithm throughout the entire searching procedure.Several classic chemical dynamic optimization problems are tested and detailed comparisons are carried out.The results illustrate that the proposed approaches outperform basic algorithms in terms of both accuracy and convergence speed.(4)To tackle constrained dynamic optimization problems(COPs)with path constraints and terminal constraints,two intelligent optimization methods combined with control parameter vectorization are further proposed.Penalty based adaptive dispersion invasive weed optimization approach(PADIWO-CVP)are able to convert the original problem into an unconstrained dynamic optimization problem by adding the constraints as additional items to the objective function.Aimed at maintaining the balance of feasible and infeasible individuals in the population,gradient based adaptive dispersion invasive weed optimization approach(GADIWO-CVP)utilizes the gradient information of constraints to correct unfeasible individuals.Three classic constraint dynamic problems for biochemical processes have been tested as illustration.The obtained results reveal that the proposed mechanism tends to not only exploit feasible regions,but also explore the boundary of feasible and infeasible regions carefully.